Re: DOS, was Re: Ada Tutor Web Site Shutting Down

[Georg Bauhaus <rm.dash-bauhaus@futureapps.de>]

On 5/12/11 8:24 PM, Adam Beneschan wrote:
> On May 12, 10:40 am, Rugxulo<rugx...@gmail.com>  wrote:
>
>>> Rather, adopting DOS seems quite
>>> frankly the most far reaching mistake that computer dependent
>>> industry has committed, the consequences being loss of both
>>> software quality and---far worse---a collective loss of any knowledge
>>> of what quality software might be!
>>
>> Not true. Quality software can be written on any OS. "A poor carpenter
>> blames his tools." Many successful things have been written for DOS,
>> not the least of which are Lotus 123, MS Word, Turbo C++, Turbo
>> Pascal, Wolfenstein 3D, Doom, Quake, and a bunch of ports of GNU
>> tools.

That's a fine fallacy.  Many successful things have been
written for X, therefore X must be good. (Think of char*.)
Even when qualities can only be attributed to the efforts of
very good carpenters tackling knot-holes and cracks in warped
material.  The "argument" omits the detail that many successful
things---and sometimes the very same things---have been written
for other OSs, too. And it omits technical scales for measuring
the qualities of technical systems such as operating systems.

At these points comparison can start, and at these points one can
actually grade OSs on a good-bad scale.

MS Word on DOS is one example.  It has qualities which I have
had to learn and to explain to a few uninitiated. It has been
a major headache for many, many users of DOS computers.
   ESC - transfer - load:
it takes a computer savvy person to actually understand and explain
what these notions are referring to.  Yet MS sold it to the masses
They wouldn't need to write software for the masses' skill level.
The masses got used to the qualities of programs like MS Word.
In part because they did not have to pay for the software.

It was expensive to train people to use these programs
effectively.

But the masses accepted the qualities of these programs.
Mastering the programs was a moderate challenge; you could compete
with your peers.  Run for MS Word mastery...

They have even built a "co-operative" multitasking graphical
Windows(TM) system on top of DOSs, yes. But suppose an Acorn Archimedes
computer had cost 1K less and that MS had not been the marketing agency
that had understood how to win cheap and sell their stuff to
office equipment suppliers.  Would you still be arguing
that DOS is the good foundation?  And RISC OS not a good foundation?

DOS was cheap.  Technically superior solutions tended to be more
expensive at the time of purchase.

Industry was already looking at different solutions when IBM,
allegedly, wanted to sell terminals that were in need of an operating
system.  Bill and Steve went to see them...
If this is true, then, obviously, industry would have chosen
differently if the choice had been by technical qualities.


> Seems right.  Actually, I find Georg's assertion puzzling.  The
> quality of software has to do with the experience of the people
> writing it, and with the design principles (or lack of) that the
> software developers apply or fail to apply.


The perception of operational qualities of an OS by its users
is important, too, as are the expectations that DOS has shaped.

Certainly programmers need gain experience with a specific OS.
They invest time in learning how to program a system. Then,
they want to build the well crafted software that Rugxulo has
listed.  Has DOS been the better foundation among the
alternatives?  Even among the simple systems?

With Tom Moran's comment in mind:

- Is DOS simple to use?  Give someone an MS-DOS manual.
If they had not had prior CS training, they will end up in
despair asking neighbors for help about what it all means.

Give some kid an iPad and watch. Granted, the system is much
more capable than a DOS PC, but most concepts are deliberately
kept simple, as simple as Microsoft Bob was to be.

- Is DOS simple to program?  INT 16#13# etc. being simple once you
get the idea does not mean that solutions built around interrupts
will be simple constructions.  I also think that this foundation of DOS
amasses a large amount of interrupt function documentation, just like
the function libraries of alternative systems do.

And every now and then, the fine DOS hiding solutions of the
Turbo brand mentioned would remind you of the rather primitive
qualities of DOS style OSs: soft rebooting the entire system when
testing just one program that was *not* the operating system.

- More than once I had to give up a floppy or hard disk whose
FAT got shot. The BIOS would help with reading out relevant
sectors; but if the dangling pointer had hit the right sector,
no BIOS would help,  A few companies were flourishing because
they provided forensic analysis and repair; had DOS had a
quality file system, PC Tools, Norton, etc. would have had
a different business plan.

But people got used to the idea that disk file storage can have
binary qualities: you can loose everything on a medium if part
of it is wrecked, or else everything is mostly fine.

A file system of different qualities means a shift in likelihood:
with more redundancy, is is more likely are that you only loose
a few files and not an entire file system.  Slightly higher
cost, higher quality file system.

> The choice of OS seems to be one of the least important
> factors of all.  It's really hard for me to imagine how the choice of
> DOS versus other operating systems available at the time (or
> potentially available) could have any significant impact on software
> quality at all.

For one thing, DOS effectively ruled out a number of design
patterns:  you need something on top of DOS in order to
operate(!) these patterns.  Such as programmable auxiliary
tasks doing background work.

I think that DOS successfully persuaded parts of the industry
that being simple implies being cheap.  But DOS PCs are more
cheap than simple.